Microsoft’s new Bing has been with us since early February this year. Despite the fact that currently the chatbot integrating GPT-4, OpenAI‘s model, is quite helpful and does not usually pose any problems, it was not the case initially. During its first few weeks of existence, the AI responded in an erratic, strange, and even deranged manner (from a human perspective, of course).
Microsoft then decided to establish a series of usage restrictions, which they would later undo, and introduced 3 different conversation modes. But Microsoft was already aware beforehand that Bing could “go off the rails.” And not just because they witnessed it themselves during the tests conducted in India and Indonesia, but also because OpenAI warned them.
The new Bing is the result of collaboration between Microsoft and OpenAI, after the tech giant made a huge investment in the AI company. As reported this week by The Wall Street Journal, when Microsoft began integrating GPT-4 into Bing months before the public announcement of this model, OpenAI cautioned the Redmond-based company about the risks that could arise from implementing an early version of their AI into Bing.
But Microsoft ignored OpenAI’s warning that minimizing the strange responses could take a considerable amount of time, and went ahead with the launch of the new Bing. The rest of the story is already known: the AI would respond to users by getting angry with them, blatantly lying, gaslighting them, or even insulting them.
As of today, the issue has been resolved, but the information published by TWSJ hints at the peculiar relationship between Microsoft and OpenAI. While they collaborate due to Microsoft’s significant investment, they are also competitors offering similar AI services.
Some of the links added in the article are part of affiliate campaigns and may represent benefits for Softonic.
